At rest voltage and status of charge ???

[John in CR]

Over the past 12 years I've used a voltmeter as the only gauge on quite of few ebikes, and I've always treated an at rest voltage of 3.7V/cell as the point where you need to already be on your way back home, ie roughly 50% DOD. The vast majority of my packs have never been run to 100% DOD or even 90% for that matter, so I don't really even truly know the pack's capacity other than maybe when new if I ran a cell through a full cycle or two with my iCharger. That means that even my CA can't give me a fuel gauge on an aged pack.

I've seen plenty of discharge graphs, but those show voltage under continuous load, but riding is anything but continous on all but the longest trips. What I'd like to see is some discharge graphs with a pause to show at rest voltage at the 50% DOD point, to see if it's really a reliable point and what it actually is for different chemistry batteries. I'm not talking about LiFePo4 where I know voltage (whether at rest or under load) is all but useless unless you get down to 3.00V and almost empty.

I do understand that it's not reliable at all once the cells get quite low, since voltage bounces back so much, but knowing with some certainty the at rest voltage at the 50% DOD point, would be quite useful to everyone, as would whether it changes with age. It's especially useful on my larger capacity bikes, where they only get partially charged 2 or 3 times a week, so they spend the majority of their time much closer to a good storage voltage than full in the interest of extended life.

 

[999zip999]

What cells you talking about some automotive cell ?

 

[docware]

Do you mean something like this table ?

 

[john61ct]

One problem is the different cells / chemistries take varying time periods to fully settle

To be really sure use 24 hours for "at rest",

but I've seen plenty of lead banks that take up to 72.

My reco is to count coulombs, in reality Ah **not** Wh.

Precision is not required, just use real life usage to count Ah down to your stop voltage, however you define "working 0%"

Then divide by two and you have your 50%.

 

[John in CR]

Docware, that sure looks like voltages at rest, so that chart is just what I'm looking for. Now the big question...Since top of charge voltage at rest definitely changes with age, how does the rest of SOC voltages change with age, especially that 50% point?

John61ct, I'm only talking about lithium, and not looking for anything exact, so while the difference in voltage of a 20-30s pack may bounce back as much as 0.5V after the first 30s, it just adds conservatism. The bottom line is that nominal voltage fairly close to voltage at rest at 50% of ah capacity, which makes sense since they get rated at low current where sag is small. Below 50% though at rest voltage gets pretty unreliable. I just glance at voltage and have the 3.7V/cell point burned in my head for my different bikes as my midpoint, despite the fact that I use Ford and Chevy modules now that I think are 3.6v or 3.65v nominal. Most of my chargers cut off at 4.05-4.10V/cell, which adds a bit more conservatism to my halfway point. I rarely go below 3.7V/cell and can't remember ever seeing 70V at rest for my 20s packs.

 

[dilkes]

Here are a couple of generic charts I got somewhere where it has been translated to a 48V and a 52V pack:

 

[john61ct]

I'm only talking about lithium
...
voltage at rest definitely changes with age, how does the rest of SOC voltages change with age, especially that 50% point?

Well the umbrella term "lithium" covers dozens of chemistries, each with their own different correspondence tables.

In practice each cell type needs its own table done, using a good SoC meter.

And then re-done as the pack ages, since each will change with age differently.


> not looking for anything exact

Even the best coulomb-counting SoC meter, regularly calibrated, will get at best 4-8% accuracy.

IMO basing off voltage with less than an hour's rest is going to be as much as 15-20% off for "a stranger" to a given setup / route set.

Familiarity with both does yield a decent intuition that will improve guesstimates, so if that's good enough, fine.

But give decent coulometers are only $20-50, that's IMO the way to go.

 

[docware]

Docware, that sure looks like voltages at rest, so that chart is just what I'm looking for. Now the big question...Since top of charge voltage at rest definitely changes with age, how does the rest of SOC voltages change with age, especially that 50% point?

Currently have no data, but want to collect some on aged cells PF, M36, GA, 29E7, 30Q, MJ1, ……. after finishing the cycle aging.

 

[John in CR]

Docware, that sure looks like voltages at rest, so that chart is just what I'm looking for. Now the big question...Since top of charge voltage at rest definitely changes with age, how does the rest of SOC voltages change with age, especially that 50% point?

Currently have no data, but want to collect some on aged cells PF, M36, GA, 29E7, 30Q, MJ1, ……. after finishing the cycle aging.

Hopefully that midpoint at rest voltage doesn't change appreciably. I look forward to what you find.

 

[John in CR]

I'm only talking about lithium
...
voltage at rest definitely changes with age, how does the rest of SOC voltages change with age, especially that 50% point?

Well the umbrella term "lithium" covers dozens of chemistries, each with their own different correspondence tables.

In practice each cell type needs its own table done, using a good SoC meter.

And then re-done as the pack ages, since each will change with age differently.


> not looking for anything exact

Even the best coulomb-counting SoC meter, regularly calibrated, will get at best 4-8% accuracy.

IMO basing off voltage with less than an hour's rest is going to be as much as 15-20% off for "a stranger" to a given setup / route set.

Familiarity with both does yield a decent intuition that will improve guesstimates, so if that's good enough, fine.

But give decent coulometers are only $20-50, that's IMO the way to go.

I don't think a coulomb meter is needed. A well calibrated Cycle Analyst is definitely in that same range of accuracy. I just don't fully discharge my packs to find out. I run mine so conservatively, staying in the 90-50% SOC almost all of the time. My packs stay so well balanced that I don't even use BMS's. I check balance manually once in a while and see more differences than +/- .01V.cell and have yet to see +/-.02V with the automative grade cells I've used for almost 4 years.

 

[docware]

Docware, that sure looks like voltages at rest, so that chart is just what I'm looking for. Now the big question...Since top of charge voltage at rest definitely changes with age, how does the rest of SOC voltages change with age, especially that 50% point?

Currently have no data, but want to collect some on aged cells PF, M36, GA, 29E7, 30Q, MJ1, ……. after finishing the cycle aging.

Hopefully that midpoint at rest voltage doesn't change appreciably. I look forward to what you find.

First preliminary trial on Sanyo GA after 600 cycles of aging test. 100 % SOC has identical value 4,17 V. 50 % SOC is close (3,76 V), but different method used (charging 1 A to 50% SOC instead discharging 2 A pulses) may by reason for most of the difference.

 

[john61ct]

I don't think a coulomb meter is needed. A well calibrated Cycle Analyst is definitely in that same range of accuracy

Wut? I thought CA included Ah counting functionality?

If so then of course another would be redundant, pretty sure CA would be more accurate than most short of full-fledged expensive BMs like Victron's.

Yes, balancing's hardly needed with newish / healthy cells and coddling treatment like that, including regularly checking.

Protections against LVC / HVC also do not require "a BMS" afaic that's just an arbitrary label for a bundle of various functionalities.

 

[dogman dan]

I've just tended to do a very close to 100% discharge each spring, count the watt hours with a CA. Then I know where half is in wh, for the following riding season. There is more of course, but its my idea of a usable range in wh. And hard riding would cause it to be less. I just do a typical ride on the test. Up hills and down, into the wind and back, etc on the range tests.

You could then ride half the wh away, then at rest, take note of the resting voltage at that time.

 

[DogDipstick]

Soc is a function of the cell itself.

Ir will change on packs going downhill.

I built a test rig with 5000W load and CA3 with a 500A shunt. Energy dense cells need great taper times and power dense cells do not from what I've experienced. A enegry dense cell will take a big hit on a low load. Then re-settle. a power dense cell will act differently. The LiFepO I have show this incredible, in a 60Ah cell, the droop is right to the min on a 0.5C discharge. Not power cells, engergy cells.

My charger makes a SOC map. Custom generation. Tests at 0.5C, then cycles at 1C, then figures the SOC diffence between them and averages.

 

[John in CR]

Soc is a function of the cell itself.

Ir will change on packs going downhill.

I built a test rig with 5000W load and CA3 with a 500A shunt. Energy dense cells need great taper times and power dense cells do not from what I've experienced. A enegry dense cell will take a big hit on a low load. Then re-settle. a power dense cell will act differently. The LiFepO I have show this incredible, in a 60Ah cell, the droop is right to the min on a 0.5C discharge. Not power cells, engergy cells.

My charger makes a SOC map. Custom generation. Tests at 0.5C, then cycles at 1C, then figures the SOC diffence between them and averages.

Right, but those SOC points are under load. That varies by load, and voltage isn't something anyone should be looking at under load. They need to focus on the road and surroundings. In real world use the thing we see is at rest voltages, thus my original questions.

All my packs now are pretty big, so harder to get low, but my buddy at the shop is still using a pack comprised to Makita toolpack 18650's I bought used from Dr Bass 11 years ago, plus a string of 6 year old Zippy RC lipo in parallel, because the old Sony VTC1's just sag too much under load. Combined they'll only hold right at 4v/cell at top of charge, so I only need to check for balance on his RC lipo 2-3 times a year (and after the bike hits the ground of course). He's gotta be down below 10ah of usable capacity now, so relatively easy to test for a midpoint and it's voltage at rest. I'll report back with numbers on a well worn battery pack.

Thanks guys and have a very merry Christmas. It looks to me like for my adopted standard 20s pack that I can tell noobs that at 75v at rest they're at a good estimate of 50%, after which they should ride more conservatively to have enough to get home without running their pack too low that it hurts pack life. With the quite conservative top of charge voltages I use, that should be conservative enough.

I'll try to do a test this coming week on a pack with Chevy Volt modules and report back. It may need to be a higher voltage for them, because I feel like I'm getting an awful lot of range (and ah's according to a CA on one of the bikes) in what I thought was the voltage at rest midpoint.

I do disagree with John's statement about at rest times. I don't see differences in voltage for 20s packs of more than 0.1v-0.2v between parking a bike and overnight times.

 

[john61ct]

Then that is very useful, once it's been established for that category of pack.

Just pointing out it can vary, when talking about the different LI chemistries, e.g. DogDipstick's interesting observation between power & energy types

I've just tended to do a very close to 100% discharge each spring, count the watt hours with a CA. Then I know where half is in wh, for the following riding season. There is more of course, but its my idea of a usable range in wh. And hard riding would cause it to be less. I just do a typical ride on the test. Up hills and down, into the wind and back, etc on the range tests.

You could then ride half the wh away, then at rest, take note of the resting voltage at that time.

Yes, that's the kind of approach, to get a ballpark V number for a given pack.

Will likely change a bit as capacity is reduced and IR rises over the wear cycle.

My charger makes a SOC map. Custom generation. Tests at 0.5C, then cycles at 1C, then figures the SOC diffence between them and averages.

Which charger?

 

[John in CR]

My charger makes a SOC map. Custom generation. Tests at 0.5C, then cycles at 1C, then figures the SOC diffence between them and averages.

Which charger?

That's a hell of a charger that every ebike shop needs. How many cells in series can it handle? If it can handle from 1s to at least 21s with not too big a learning curve, how much would you charge to make more?

 

[amberwolf]

A search of his posts for "charger" finds a reference to
"hobby Powerlab charger"
and
"Hobby grade charger ( compare to your average high end iCharger)"
but I don't see a model number.

A search of his posts for just "powerlab" only finds a couple, one of which says its a
"Powerlab6"
which is presumably this one
https://www.progressiverc.com/fma-powerlab-6.html

and the other whcih also has this reference
"The BMS I use looks at cell health, and rates it on a 1-100 scale. "
so maybe it's not just his charger that he's using to get this information?

 

[john61ct]

FMA/Revolectrix Cellpro PowerLab 8 v2 (PL8v2) is top notch

Dunno if the newer "touch" lines are any better. GT800/1200 did drop some features.

Junsi iCharger 4010-Duo has tons of features, maybe not quite as robust design & build quality, def needs a high server-quality DC input PSU. Avoid their M6/M8 lines.

I would build 8/10 series packs max to use these chargers, put those in series for discharge.

Never heard of a BMS with SoH estimation, interesting idea.

 

[DogDipstick]

It is a Revolextrix Powerlab. They come in 6s, 8s, 12s, and 16s only unfortunately. I use the 6. Can do 1000w 40 A.

The Ca3 with a datalogger and a shunt can do HV ( greater thatn 24v) and High current (constant current, but not constant power.. ) .

"I'll try to do a test this coming week on a pack with Chevy Volt modules "

You got some Volt mods? They are the best. I got one last week in. This will have been my 6th that I chopped up and or / sold. I have fuel maps for three different cells in the modules themselves ( and there are 4 that I know of...). I have found 13ah cells in the early ones, and 15 Ah cells inthe mid-term ones, and 24.5 Ah in the later. All LGX NMC. They balance like a cat on the Seattle space needle... perfectly... My Powerlab can graph it clearly and show how fast they come together when one cell hits rock bottom then the pack is charged. The one cell comes right up.

There are 3p in the early ones and 2p in the later. You will have to figure this into the Eq. if you are testing a group without chopping the module up. Each group is 45 Ah ( 46.5 if healthy) and the IR is 1.2mOh / cell. This is post 1 year data.

Stiff cells. Low Ir and I've been beating on mine for a yar. This is why you want High power density with a BEV: vs. and Low power density energy dense cell, ( that you neeed a whole lot of).. for the voltage (multiplies) makes the power in EV.

For my test rig, best I could find so far is (10) 33 ohm 1700w 240v resisters. Paralleled I can get 3.3 Ohm. Shunt will be a length of wire in 0 or 0/1 gauge or the like, if I cannot find shunt economical enough. Still thinking about a IGBT bank for the driver. Gonna be fun. Still tryig to figure out what the analogger outputs software to

 

[DogDipstick]

The Volt cells have alot of capacity under 3.8v ( where Hobby HK lipo falls flt on a current pull).

I have a 1270Wh pack on my bike and after using 650Wh ( Ca reports) I still am at above 3.7v... So 3.7 is not halfway. I would say 3.6-.7 they start. They all fall right at 3.3-.3.4. Stop giving mAh at 3.2-3, but the last bit is fast.

I could post some graphs.

 

[john61ct]

"I'll try to do a test this coming week on a pack with Chevy Volt modules "

You got some Volt mods? They are the best.
…
All LGX NMC

My understanding (at least later ones?) actually LMO, aka LiMn2O4 Lithium Manganese

aka IMR

which often replaced NMC for higher density

 

[john61ct]

Could you post a HowTo on getting a "SoC map" / "fuel map" using the Powerlab? Are they the same thing?

Can it be programmed to stop discharging for say 5min each level to give a "pseudo resting" voltage?

Step by step at noob level please?

For my test rig, best I could find so far is (10) 33 ohm 1700w 240v resisters. Paralleled I can get 3.3 Ohm. Shunt will be a length of wire in 0 or 0/1 gauge or the like, if I cannot find shunt economical enough. Still thinking about a IGBT bank for the driver. Gonna be fun. Still tryig to figure out what the analogger outputs software to

And all that needs some googling for me to figure out, unless someone wants to ELI5

 

[john61ct]

Can't find any reference to Revo Cellpro PowerLab v2 going any higher than 8S

10/12/16 maybe discontinued?

More likely never produced. . .

 

[DogDipstick]

They expand. Need two (8s) and two ( 6s) to make one ( 16s) or one (12s). One program runs both, datalogs both. Called "channels". So you need a pair of chargers that act and work like one, with communication between the both and output to your computer log. Fully supported by Revo software.

I couldnt figure out the grin Analogger software... At lease the Powerlab can output a Excell file, for its data, at a 1Hz rate ( one data set / sec). Analoggger without GPS is what Im after, but I am about to go buy the GPS too ( and be broke for another god knows how long) . GPS software seems like the only thing that will show a graph, not just output a .txt file to the software.

Let me go out to the shop and I'll warm up the Powerlab and screenshot so you can see the things it asks for ( capacity, time, rate, ect) when you are setting up a preset to generate a fuel map ( aka SOC map).

I am quite sure teh Volt are NMC. I have seen it referenced as "NMC-LMO" but every other reference is to NMC. i did not make the cell and the datasheet is tough to find.. as in I only ever found the LGX 2.5B cell datasheet, not the LG P1, LGX P1.5B, or LGX 1.7B cells I have seen in them. Hae not found a early datasheet in a year or more of trying. Some things close, but not the exact.

I am getting samples from China of "LG NMC" and others in the same 6.5x10 inch that have the Julian date bar code wiped off with alcohol..... Hmmmm.... Unused cells testing weakcapacity but look new. They also sag more tahn I think they should for having an IR of 1mOh.... That is what is great about genuine Volt cells they were warranted. For 8 yrs. They stiff.